Inter-American
Development Bank
Science and Technology
Division, Social Sector.
TECHNICAL NOTES
No. IDB-TN-319
Employment Generation,
Firm Size, and Innovation
in Chile
Roberto Alvarez
José Miguel Benavente
Rolando Campusano
Conrado Cuevas
October 2011
Employment Generation, Firm Size,
and Innovation in Chile
Roberto Alvarez
José Miguel Benavente
Rolando Campusano
Conrado Cuevas
Inter-American Development Bank
2011
http://www.iadb.org
The Inter-American Development BankTechnical Notes encompass a wide range of best practices, project
evaluations, lessons learned, case studies, methodological notes, and other documents of a technical
nature. The information and opinions presented in these publications are entirely those of the author(s),
and no endorsement by the Inter-American Development Bank, its Board of Executive Directors, or the
countries they represent is expressed or implied.
This paper may be freely reproduced.
Employment Generation, Firm Size, and Innovation in Chile
Roberto Alvarez
José Miguel Benavente
Rolando Campusano
Conrado Cuevas
Centro INTELIS
Universidad de Chile
*
Abstract
This paper compiles and analyzes several sources of information to shed light on the relationship
between innovation and employment growth in the manufacturing industry in Chile in the last 15
years. Our overall conclusions are that process innovation is generally not found to be a relevant
determinant of employment growth, and that product innovation is usually positively associated
with an expansion in employment. These results seem to be similar regardless of firm size and
hold for both low- and high-tech industries. Our findings reveal that in-house R&D (make only
strategy) is positively related with employment growth. However, this is not the case for the
sample of small firms. With respect to employment growth by types of workers, skilled vs.
unskilled, the evidence is less robust. However, in-house R&D appears to favor employment
growth for both types of workers in low-tech industries.
JEL Classification: D2, J23, L1, O31, O33.
Keywords: Innovation, employment.
*
The authors would like to thank the IDB team, external reviewers, and participants at MEIDE 2011 and
internal seminars at INTELIS, University of Chile, for their valuable comments and suggestions.
1. Introduction
Since the mid-1990s, successive governments in Chile have promoted innovation policies
aimed at improving long-term productivity and economic growth. However, until now,
neither an impact evaluation of these particular efforts nor a dynamic analysis of innovation
as a whole has been carried out for analyzing the potential effects of innovation on
employment.1 This is a relevant empirical issue considering that theoretical considerations
do not generate unambiguous predictions for this relationship. Moreover, the empirical
evidence shows mixed results regarding the impact of innovation on employment growth,
especially process innovation (Harrison et al., 2008).
The purpose of this paper is to analyze empirically the effects of innovation on
employment growth using microeconomic data for Chile. This exercise is relevant for
understanding, among other things, whether process and product innovation create or
displace employment, how different innovation strategies affect employment, and how
these effects vary by firm size and sectoral technological intensity. Several studies have
been conducted on this issue, but they apply mostly to developed countries (Pianta 2005).
In this context, the present study has the following specific objectives: (i) to assess how
different types of innovations create or displace employment in Chilean firms (focusing on
the differential effects of product and process innovations on employment growth); (ii) to
look at how different types of business innovation strategies influence the capacity of
innovation to generate or destroy employment, and (iii) to determine whether the types of
innovation (product or process) and innovation strategies have differential impacts across
firms of different sizes and between high- and low-tech sectors. To analyze (ii), we
consider two types of innovation strategies: make (in-house developed innovation) and buy
(externally acquired innovation). Specifically, we analyze the impact of innovation on
employment growth for small firms and (iv) how different types of innovations and
innovation strategies affect the quality of employment as measured by skilled and unskilled
employment growth. The differential impact for small firms is closely assessed.
There are important features of the Chilean economy that justify a study of this type.
First, there are several available versions of the innovation surveys that follow the
1
One exception is the study carried out by Benavente and Lauterbach (2008) estimating the impact of product
and process innovation on firm employment growth using the 2001 Chilean innovation survey.
2
Community Innovation Survey (CIS) methodology (Oslo Manual) and allow a comparison
of our results with other studies using similar surveys. This paper, uses mainly the
innovation surveys carried out in 1995, 1998, 2001, and 2007. This information is
complemented by firm-specific information obtained from the Annual Survey of
Manufacturers (ENIA). This link between these sources of information is relevant given
that innovation surveys usually provided very limited information on firm characteristics
such as employment, sales, exports, and investment. Unfortunately, the same information
for analyzing this issue in other sectors, such as services, is not available.2
Second, in recent decades, the Chilean government has implemented several policies
aimed at increasing innovation and productivity. If innovation affects employment, these
policies may have important effects on workers’ welfare. Finally, as has been widely
documented, since the financial crisis at the end of the 1990s, the Chilean economy has
experienced a significant slowdown in productivity. After a period of high economic
growth between 1986 and 1997, the rate of Chile’s aggregate productivity growth declined
dramatically in the last 10 years (Fuentes, et al., 2006). There is no consensus among
policymakers and researchers on the main causes of this phenomenon. This study may shed
light on the relationship between innovation, employment, and productivity.
A number of empirical analyses have examined the determinants of innovation using
different versions of the innovation surveys carried out in Chile since 1995. Crespi and
Katz (1999) and Crespi (1999) analyzed how industry and plant characteristics might
explain differences in innovation using the first version of the survey. Benavente (2005)
extended that analysis using three versions of the surveys. Álvarez (2001) and Álvarez and
Robertson (2004) focused on trade-related variables as the main drivers of innovation
activity. More recently, Benavente (2006), Álvarez et al. (2010), and Crespi and Zuniga
(2010) estimated the impact of innovation on firm productivity. The study undertaken by
Benavente and Lauterbach (2008) is the only one that analyzes the relationship between
innovation and employment growth.
This paper extends the work of Benavente and Lauterbach (2008) in several
dimensions, using additional innovation surveys and including new empirical questions.
These authors found that product innovations affect employment positively and
2
Since the 2001 Survey, there are additional sector included in the sample and not only manufacturing firms.
3
significantly, but process innovations do not. However, they did not look at the effect of
innovation on employment composition (skilled versus unskilled). Nor did they analyze the
effect of different innovation strategies on employment growth, the existence of potential
differences between small and large firms, or technological intensity across sectors
Moreover, their paper focuses on innovative firms only. This paper uses a different
identification strategy that enables the full sample of firms to be used.
This paper is structured in the following way. The second section presents some
qualitative and quantitative information about the Chilean economy, innovation policies,
and the relationship between innovation and employment. It includes a summary of
interviews and cases studies analyzing how innovation and employment are related in
specific Chilean firms. The third section describes the datasets used in this paper. The
fourth section presents the methodology and econometric results for the relationship
between innovation and employment quantity. The fifth section focuses on the empirical
analysis of innovation and employment quality, using estimations for skilled and unskilled
workers. Section 6 presents the analysis of the effect of different business strategies on
employment growth. Section 7 provides the main conclusions and findings. The paper
concludes with policy implications.
2. The Chilean Economy
2.1 General Description
Chile has enjoyed remarkable economic performance in the last 15 years. With an average
annual rate of growth slightly over 4 percent and annual population growth nearing 0.9
percent, today GDP per capita is close to US$13,000 (PPP). Some relevant variables of the
Chilean economy and the context of innovation are summarized in Table 1.
Despite their volatility, unemployment on average has never reached two-digit figures.
Last year, despite the effects of the financial crisis and the earthquake, the unemployment
rate increased to almost 8 percent, only slightly higher than the 6.8 percent in 2005.
Nevertheless, there is some consensus that the quality of human capital is relatively low
and needs to be improved drastically in the years to come (Brandt, 2010). As can be seen in
Table 1, only a quarter of the total labor force has tertiary education. If innovation—which
is heavily based on well-trained employees—is at the core of the country’s long-run
4
economic growth strategy, it seems that the quality of human capital will, at some point, if
not now, be an obstacle for maintaining high and sustained productivity growth.
Table 1. Evolution of Main Variables in Relation to Employment and Innovation
Indicator
1995
2000
GDP per capita (US$ PPP) (1)
9,140
10,470
-
4.2%
4.2%
4.2%
-
2.9%
1.5%
1.6%
Share of the population in the labor force (4)
39.1%
39.4%
41.5%
45.7%
Share of tertiary educated as a proportion of the labor force (5)
20.9%
22.8%
25.2%
25.2%
Unemployment (6)
4.7%
8.3%
6.9%
7.8%
Unemployment of the tertiary educated (7)
7.0%
6.4%
6.5%
9.2%
Innovation/GDP (8)
1.3%
1.0%
0.8%
0.9%
R&D/GDP (9)*
-
0.1%
0.2%
0.4%
BERD/GDP (10)*
-
0.1%
0.2%
0.2%
GDP growth (between periods) (2)
Labor productivity growth (growth between period averages)
(3)
2005
2010
12,172 13,057
Notes: (1): constant 2005 international $; most recent: 2009; Source: World Bank. (2): growth of constant 2000 US$ GDP;
most recent: 2009; Source: World Bank. (3): labor productivity = gdp/employment; most recent year: 2009; source:
Central Bank of Chile. (4): most recent: 2008; Source: World Bank.(5): Labor force with tertiary education (% of total),
most recent: 2007; source: World Bank. (6): most recent: 2008; source: World Bank.(7): Among those with tertiary
education, unemployed/labor force; Source Casen 1998, 2000, 2006, 2009. (8): most recent: 2006: source: national
innovation survey. (9), (10): most recent 2009; source: national innovation survey. * Year 1995 it’s not informed do to a
change in the methodology.
One important factor for enhancing economic growth is the amount of resources
invested in developing and applying new ideas, products, and productive processes. Figures
in Table 1 clearly show that expenditures in innovation and in R&D are not that high
compared to other countries that have moved from middle-income to developed. Official
figures show that R&D as a percentage of GDP was 0.4 percent in 2010 where private
effort represents less than half of national expenditure.3 Moreover, this low level of R&D
investment is lower that would be expected according to Chilean per capita income. Several
studies have shown that Chile suffers an innovation shortfall (Kharas, 2008; Maloney and
Rodriguez-Clare, 2007). Maloney and Rodriguez-Clare (2007) show that even correcting
for the structure of the economy, Chile invests relatively less in R&D than its counterparts
in advanced economies.
3
In 2010, a major methodological change was implemented in order to measure R&D at the private and also at
the university level. The figures were then updated using this OECD accepted methodology back to 2000.
This explains the overestimation shown for 1995 in Table 1.
5
2.2 Policy Landscape and Recent Evolution
Since Chile was accepted as member of the OECD group, it not only has instituted new
procedures for calculating innovation and employment figures, but has also reorganized the
institutional setup aimed at supporting R&D and innovation. Most of the implementation
agencies related to innovation, R&D, education, and training fall under different ministries,
as shown in Table 2. A Ministerial Commission on Innovation (CMI), established in 2010,
has been attempting to resolve some of the coordination problems. These include dealing
with program duplication, aggregating and consolidating budgeting processes, and
designing new innovation-related programs. This is a new approach to decision making at
the government level and a dramatic departure from the practice of previous
administrations. The role played by the Ministry of Finance is far weaker than before, while
the Ministry of Economy has taken the lead in this process.
Table 2 shows that the linkages among the different units especially related to
innovation and employment are still weakly connected and there is plenty of room for
improvement. By the same token, private institutions represented by firms, unions, and
other collective agents, as well as training programs, do not actively participate in the
decision making process associated with R&D and innovation efforts. This needs to be
remedied in the Chilean case, as was revealed in some of the interviews of firm managers
conducted for this study.
6
Table 2. Innovation and Employment Policies
Innovation
CORFO
National
Interactions
and
consultation
mechanisms
with other
considered
dimensions
(4 point scale
(a)
)
2
Employment
policies (emphasis
on R&D, skilled
employment)
Ministry of Labor,
CONICYT/MIDEPLAN
National
2
2
1
Higher education
(universities,
vocational
education)
Ministry of Education, CONICYT
National
2
2
1
Training policies
(on the job,
lifelong learning,
etc.)
Ministry of Labor, SENCE
National
2
2
2
Dimension
Responsible agencies
Level of
influence
Privatepublic
partnerships.
Participation
of the private
sector
(4 point
scale(a))
Policy
assessment
mechanisms
2
2
(4 point
scale(b))
Source: Authors elaboration based on interviews.
Notes: (a) Scale: 1: Non-existent; 2: Low; 3: High; 4: Very high. (b) Scale: 1: Non-existent, 2: Some assessment (non compulsory) exist;
3: Assessments are inherent part but the results are not necessary taken into account in the (re)design of the program; 4: Assessments are
made and results imply redesign/abandon of programs.
The National Council of Innovation for Competitiveness has played a major role in
suggesting and promoting a more efficient institutional setup designed to improve the
performance of the Chilean National Innovation System. It promoted the creation of CMI,
and endorsed an aggregate budgeting process among all agencies involved in promoting
science, technology, innovation, and entrepreneurship in the country. However, there are
still several issues related to innovation and employment (and to other issues such as
science, trade, and intellectual property rights) that need to be addressed. For example, with
respect to upgrading the skills of the labor force, which can be considered a relevant
constraint for innovation and productivity growth, even though there is awareness of this
problem, the challenge is to create the institutional setup to solve the coordination problem.
Some other challenges and potential actions to be taken are summarized in Table 3.
7
Table 3. Challenges, Strengths and Actions in Relation to Innovation, Employment
Creation, and the Upgrading of Skills of the Labor Force
Dimension
Main challenges faced by
the country
Innovation
-Low number of firms
performing
routinized
R&D
and
innovation
activities.
-Weak interaction among
main players involved in
innovation.
-Huge
gap
between
technical feasibility and
commercial viability in
thinking and implementing
innovation projects.
-Several
bottlenecks
surrounding
innovation
activities that need public
intervention to solve them.
Employment
creation
- Labor quality rather than
quantity is becoming a
major issue in Chile.
However, highly
heterogeneous patterns
across sectors.
- Turnover rates have
increased in recent years.
- Exchange rate
appreciation may affect
tradable and non tradable
sectors differently (Dutch
disease)
-Shortage of graduates in
S&T especially for natural
resources based sectors.
Upgrading of
the labor
force skills
Challenges at the
institutional/policy
dimension
-No further support
towards selectivity and
regional issues related
with innovation.
-Weaker role of the
National
Innovation
Council.
-Several coordination
failures
at
the
government
level
(isolated efforts)
Main strengths
Actions taken/ to be
taken
-Consensus
about
innovation and
entrepreneurship
as a key solution
for
economic
growth.
-Few but highly
visible
successful
innovation
stories.
-Implementation
of
new
programs that
emphasizes
demonstration
effects (start-up)
-Modification of the
R&D tax credit law
allowing in house
R&D activities to be
considered
for
benefits
-Strengthen
demonstration
effects.
-Institutional
refinements
(Millennium
Initiative* now under
the
Ministry
of
Economy; a similar
situation may happen
with CONICYT)
-Labor market rigidities
partly explain low TFP
growth.
-Discussion about
labour markets issues
are not related with
productivity.
-Low interaction of
Ministry of Labour and
other ministries.
-Job creation
likely to persist
in the next two
years
- Awareness that
job quality
needs more
attention and
innovation
needs better
trained workers.
- More resources
allocated to training
and competences
certification
programs.
-Start up Chile may
have an impact on
new jobs most of
them for highly
qualified Chileans.
-Lack of information
(and those responsible
to generate it) about
needs of human
resources at a sectoral
level.
-Disconnection between
those who supply
labour skills upgrading
and the potential
demand for them
(information
asymmetries)
-Awareness
(public and
private sector)
of the existence
of coordination
problems.
-Creation of ad hoc
councils to discuss
ways to solve the
coordination
problems (at the
government level and
at the market one)
Source: Authors’ elaboration based on interviews and previous work.
Notes:* The Millennium Science Initiative (MSI) is a government entity whose main objective is to promote the development of cuttingedge scientific and technological research in the country, as a key factor for sustainable, long-term economic and social development.
The MSI finances the creation and development of Centers of Research -Millennium Institutes and Nuclei- which are selected through
public grant competitions for their scientific merits.
8
2.3 Views on the Relationship between Innovation and Employment
To complement the quantitative work shown in the following sections, we conducted
several interviews with researchers, policymakers in relevant public agencies, and firm
managers. The first part of the interview described the project and its main objectives.
Interviewees were asked about their views regarding the relationship between innovation
and employment, the main obstacles to innovation in Chile, the role of labor regulations
and skill composition, and the differential impact of innovation strategies on employment.
In the case of policymakers, the interviewees were asked whether or not public policies on
innovation take into account the potential effect of these policies on employment. These
interviews were applied to two researchers in academic institutions in Chile4 and two
highly positioned members of public agencies: Ministry of Economy and CORFO.5
The main findings from these interviews are summarized below:
•
Consistent with the theoretical literature, there is some consensus that the effect of
process and product innovation on employment might be different. All of them
mention that the most likely effect of process innovation is to reduce labor demand and
that product innovation—by increasing the demand for goods produced by the firm—
has a positive effect on employment. However, the negative effects of process
innovation are considered more relevant in the short run. Particularly people involved
in public policy making believe that innovation—in a broad sense—has positive effects
on firm competiveness and then on employment creation. Both of those interviewed
are of the view that innovation—through the creation of new companies—favors
employment growth in the economy.
•
From these interviews, in general, it can be inferred that the effects of innovation on
employment composition are considered important, but not necessarily from the
standpoint of the formal skills of workers. Although academic researchers are
influenced by ideas about skill-biased technological change, policymakers mention
other relevant characteristics such as tolerance of uncertainty and capacity of
integrating knowledge. In some cases, the effects of innovation on organizational
4
5
Andrés Zahler (Universidad Diego Portales) and Daniel Goya (Universidad Católica de Valparaíso).
Conrad von Igel (Ministry of Economy) and Cristobal Undurraga (CORFO)
9
structures are also thought to be important for thinking about how firms manage the
consequences of innovation-driven changes within them.
•
The potential negative effects on employment are not considered to be important
obstacles to innovation. One aspect that was mentioned by all of those interviewed is
related to cultural issues, for example the absence of appetite among entrepreneurs of
small firms for undertaking innovation, conformity in maintaining a low-profitability
business, the high social and financial costs of undertaking unsuccessful projects, and
college graduates who are highly qualified but not interested in starting their own
business.
•
There is some agreement that labor regulations are not a significant obstacle to
innovation, but the excess of bureaucracy and the current bankruptcy law are seen as
serious impediments to the creation of new business. For policymakers, the existence
of restrictions on foreign workers and part-time jobs are also considered to be factors
that inhibit innovation in Chile.
•
Regarding innovation strategies—make or buy—there is no uniform view about their
consequences. This may be a reflection of the fact that the low levels of innovation in
Chile and expenditure in R&D highly concentrated in large firms do not permit them to
draw any definitive conclusions about this phenomenon.
•
No evidence that innovation policy has had any specific effects on employment growth
in Chile was found. In fact, the common view is that innovation increases firm
competiveness in the long run and that it creates more jobs that those that are
destroyed.
These views are generally consistent with those of firm managers in different sectors of
the Chilean economy. We have chosen four firms to illustrate in detail the relationship
between innovation and employment. Two of them were chosen using the following
criteria. First, they are small firms with fewer than 50 workers. Second, they do not
produce in the traditional export sectors of the economy. Personal contacts were used to
contact the managers. In these first two cases, both managers graduated from the
University of Chile and have more than five years of managerial experience. In the other
two cases, we looked at firm data and contact information contained in the innovation
surveys using two criteria. First, one of them should belong to a more traditional export
10
sector and the other should be a producer in a domestic-oriented industry. In addition, we
sought firms that had innovated in one of the aspects considered in the survey (product,
process, and organizational innovation).
The interview format was developed and sent to each person to be interviewed two
days prior to the interview. The case studies, presented in the Appendix, were structured
around the following issues: main innovations undertaken by these firms, the relationship
between innovation and employment, the obstacles to innovation, and views about
innovation policies in Chile. Following are the main findings of these cases studies:
• Innovation does not appear to be associated with dramatic changes in employment
in these firms, although the managers consider that the effect of innovation on
employment creation or displacement depends on the type of innovation. Consistent
with the view of researchers and policymakers, product innovation is considered to
be more likely to create new jobs, while process innovation tends to displace
workers, especially unskilled labor.
• However, this negative effect of process innovation can be minimized through
worker training. In this context, some managers had a negative impression of the
current training programs in the Chilean economy. This should be taken into
account by government programs aimed at fostering innovation.
• There are still significant obstacles to innovation in the firms surveyed. Some of
them are related to the lack of resources for undertaking innovation projects, while
others are more linked to scant experience in taking advantage of government
assistance programs. None of the managers interviewed, however, felt that the
potential negative effect of innovation on employment has been a significant
impediment to incorporating new productive processes. In fact, competitive
pressures have spurred the incorporation of new processes or the development of
new products.
In sum, researchers, policymakers, and managers believe that the effects of product
innovation on employment growth are positive. There are some concerns, especially in the
short run, regarding the potential negative effects of process innovation. However, it seems
that once firms decide to introduce new products and processes, or policymakers implement
new programs for fostering innovation, the potentially negative effects on employment are
11
not considered to be of first-order importance. The following sections provide a more
systematic analysis of the relationship between innovation and employment using a large
sample of firms over the last decade in Chile. The next section describes the data used in
this paper.
3. Data Description
The main source of information on innovation activities in Chile is the National Innovation
Survey (EIT) carried out by the National Institute of Statistics. The survey has been
conducted every three or four years since 1995. The questionnaire follows the guidelines of
the Oslo Manual developed by the OECD (OECD, 1997). Although there has been some
variation in the number and types of questions, the structure of the survey has remained
similar across the different waves.
The questions are divided among the following sections: (i) the types of innovations
that the firm has carried out in the past two or three years, (ii) the goals of those
innovations, (iii) sources of the ideas of innovation, (iv) purchases of equipments, (v)
obstacles to innovation, (vi) links with scientific and technological institutions, (vii)
importance of innovation in firm business, (viii) cost and financing of innovation, (ix)
expenditure in R&D, and (x) perspectives concerning future innovations.
We have access to four waves of the surveys—1995, 1998, 2001 and 2007—which
have been matched with information from the National Annual Manufacturing Survey
(ENIA) using a plant identification number. Due to problems originating in this
identification number between the Innovation Survey carried out in 2005 and the ENIA, it
was not possible to find a reliable matching between these two sources of information for
that year.
Table 4 presents the descriptive statistics of the sample used in this paper. The total
number of observations is 2049, with an average number of workers per plant of 215.
Thirteen percent of the plants are foreign, and more than 50 percent of the plants are
located in the Metropolitan Region of Santiago. Regarding innovation performance, about
30 percent of the plants report that they do not innovate either in product or in process, and
53 percent of them are product innovators. Note that more than 50 percent of these plants
that introduce product innovations are also process innovators.
12
Table 4. Descriptive Statistics for all Manufacturing, 1993–2006
Mean
Median St. Dev.
Min.
Max
2049
Number of observations
Distribution of firms (%)
Noninnovators (no process or product innovations)
30.70
Process only innovators (non-product innovators)
3.95
Product innovators
53.44
(of which product and process innovators)
51.05
Number of employees at the beginning of (each) survey
215
111
292
10
3625
Foreign ownership (10% or more)
0.13
0
0.33
0
1
Located in the capital of the country
0.52
1
0.50
0
1
All firms
-0.18
0.00
14.65
-67.43
62.14
Noninnovators (no process or product innovations)
0.80
0.00
14.59
-66.79
58.32
Process only innovators (non-product innovators)
2.12
3.62
14.02
-41.01
34.01
Product innovators
-0.52
-0.39
13.89
-67.43
58.59
7.59
7.88
16.27
-127.22 112.61
All firms
6.50
6.90
16.91
-66.06
66.66
Noninnovators (no process or product innovations)
2.89
3.63
18.36
-66.06
58.16
Process only innovators (non-product innovators)
7.05
7.09
14.81
-40.00
43.56
Product innovators
8.49
8.63
15.51
-47.97
66.66
Old products
-5.69
-4.17
17.51
-48.08
51.18
New products
14.2
10.5
14.0
0.1
64.3
All firms
6.68
6.47
18.36
-83.52
96.46
Noninnovators (no process or product innovations)
2.09
2.51
18.49
-83.52
59.69
Process only innovators (non-product innovators)
4.93
3.37
16.55
-27.93
63.53
Product innovators
9.00
8.42
16.83
-64.87
81.70
All firms
5.01
3.50
6.60
-10.87
36.38
Noninnovators (no process or product innovations)
3.81
2.91
5.50
-3.00
36.38
Process only innovators (non-product innovators)
5.03
2.91
8.18
-2.39
35.66
Employment growth (%) (yearly rate)
Growth wage bill per worker (%) (yearly rate)
Sales growth (%) (nominal growth) (yearly rate)
of which:
Labor productivity growth (%) (yearly rate)
Prices growth (%)
5.62
4.09
6.58
-10.87 36.38
Product innovators
Source: Authors’ elaboration based on Innovation Surveys.
Notes: Sales growth for each type of firm is the average of variable g and averages for old and new products are the
averages of variables g1 and g2, respectively. Prices computed for a set of industries and assigned to firms according to
their activity
13
During this period, the average plant increased nominal sales by 6.5 percent (more than
1 percent in real terms approximately) and employment contracted by 0.2 percent.
Employment and sales growth was higher for process only innovators (2.5 percent and 8.8
percent, respectively). While product innovators expanded sales by 8.2 percent, they
experienced an average employment contraction of 0.5 percent. In terms of labor
productivity growth, the data show an average increase of 6.7 percent. The growth rate was
lower for noninnovators (2.0 percent) than the other groups, and larger for organizational
change and product innovators, with growth rates of 9.7 percent and 8.7 percent,
respectively.
Table 5 shows similar information for the sample of small firms, defined as employing
50 or fewer workers. The total number of observations is 652, with an average number of
workers per plant of 26. The percentage of foreign plants (6.0 percent) is lower than the
same figure for all firms (13.0 percent). Similar to the entire sample, about 50 percent of
the plants are located in the capital of the country.
In terms of innovation variables, about 57 percent of the plants do not innovate either
in product or in process. This is a larger figure compared with the sample of all firms. The
percentage of product innovators is 34.2 percent, lower than the figure for all firms. In
general, the comparison between both samples shows that small firms tend to innovate less
than large firms.
The average plant employment growth for small firms was 1.5 percent, and nominal
sales increased by 5.3 percent. Process only innovators experienced the highest increase in
sales (9.1 percent), followed by organizational change innovators (8.2 percent) and product
innovators (8.2 percent). The process only innovators were also the group of plants having
the highest increase in employment growth, well above the rest of the firms.
There were also significant differences in labor productivity growth across firm size
and types of innovators. Compared to the growth of the entire sample (6.7 percent),
productivity in small firms only grew 3.9 percent. Among small plants, the highest labor
productivity growth occurred in the organizational change innovators (9.8 percent),
followed by the product innovators (6.7 percent). Similar to the case for all plants, the
lowest productivity growth in small plants corresponded to noninnovators (1.5 percent).
14
Table 5. Descriptive Statistics for Small Firms in Manufacturing, 1993–2006
Mean Median St. Dev.
652
Number of observations
Min.
Max
Distribution of firms (%)
Noninnovators (no process or product innovations)
57.52
Process only innovators (non-product innovators)
3.68
Product innovators
28.83
(of which product and process innovators)
25.92
Number of employees at the beginning of (each) survey
26.09
24
11.29
10
50
Foreign ownership (10% or more)
0.06
0
0.23
0
1
Located in the capital of the country
0.48
0
0.50
0
1
All firms
1.45
0.00
15.22
-53.39 58.32
Noninnovators (no process or product innovations)
1.49
0.00
14.62
-49.04 58.32
Process only innovators (non-product innovators)
4.54
3.79
11.05
-13.88 34.01
Product innovators
1.83
0.00
14.96
-44.12 52.23
6.47
6.34
15.21
-67.67 95.44
All firms
5.30
5.52
18.71
-65.94 63.93
Noninnovators (no process or product innovations)
3.08
4.18
18.86
-65.94 58.16
Process only innovators (non-product innovators)
7.61
9.02
11.14
-23.48 27.35
Product innovators
8.46
7.27
18.34
-43.39 63.93
Old products
-5.75
-4.42
18.77
-46.75 38.51
New products
14.21
10.15
14.58
All firms
3.86
3.90
19.97
-83.52 96.46
Noninnovators (no process or product innovations)
1.58
2.22
19.17
-83.52 59.69
Process only innovators (non-product innovators)
3.07
5.85
10.97
-16.18 22.06
Product innovators
6.63
7.38
19.26
-64.87 68.85
All firms
3.62
3.85
18.85
-83.52 75.00
Noninnovators (no process or product innovations)
1.60
2.22
18.40
-83.52 59.69
Process only innovators (non-product innovators)
3.27
0.80
10.56
-11.89 22.06
6.43
7.38
18.12
Employment growth (%) (yearly rate)
Growth wage bill per worker (%) (yearly rate)
Sales growth (%) (nominal growth) (yearly rate)
of which:
0.51
63.08
Labor productivity growth (%) (yearly rate)
Prices growth (%)
-64.87 64.36
Product innovators
Source: Authors’ elaboration based on Innovation Surveys.
Notes: Sales growth for each type of firm is the average of variable g and averages for old and new products are the
averages of variables g1 and g2, respectively. Prices computed for a set of industries and assigned to firms according to
their activity.
15
4. Innovation and Employment Quantity
4.1 Methodology
Since the literature suggests that the effect of innovation on employment generation
depends on the relative intensity of displacement and compensation effects that may
influence the demand for labor (Dachs and Peters, 2011; Harrison et.al, 2008; Hall et al.,
2008; Peters, 2005), this section decomposes both effects. Hence, the basic model
distinguishes two types of products, older and new. Thus, the change in employment can be
divided into the part due to improving efficiency related to process and organizational
innovation and the part due to the introduction of new products (product innovation). The
quantitative analysis of innovation effects on employment follows the methodology
developed by Harrison, et al. (2008) and applied in several other studies (Dachs and Peters,
2011; Hall et al., 2008; Peters, 2005). First, the following basic equation is estimated:
l = á 0 + á1 d + ! 2 g + õ
where l is total employment growth, g is real sales growth, and d is a dummy variable equal
to 1 if the firm introduced either process or product innovations. Industry- and year-specific
effects are included to control for unobserved heterogeneity at the industry level and
common shocks to all firms. The model is also estimated including a dummy variable for
product innovation and other for process innovation. As control variables, a dummy for
firms located at the capital of the country and another dummy for foreign owned firms are
introduced.
An initial problem with this approach is related to the fact that sales of new and old
products are not directly observed in the data. Following Harrison et al. (2008), these
variables can be computed based on information about (i) the share of new products in total
sales, and (ii) the growth rate of total sales. More formally, the sales growth of new (g2) and
sale growth of old products (g1) can be defined according to the following expressions:
g 2 = s (1 + g )
and
g1 = g ! s (1 + g )
where g is the growth rate of total sales and s is the share of new products in total sales.6
6
All nominal variables were transformed to real ones using a 3-digit deflator developed by Bergoeing et al.
(2005). For the last years of the sample, these deflators were updated using the same methodology and data
sources of these authors.
16
The initial and final period for computing growth rates depends on the way that
questions about innovation have been designed in the different Chilean surveys. In the first
three innovations surveys, the questions were related to the last three years. In the last
available survey, these were related to the last two years. Employment and sales growth are
computed as the annual rate of change to eliminate potential problems associated with
differences in the timing of the innovation.
Unfortunately, the Chilean innovation surveys did not ask for the exact value of s;
rather, the firms were asked about the percentage of total sales corresponding to innovated
(new) products using five intervals: 0, 0–10 percent, 11–30 percent, 31–70 percent, and 71–
100 percent. The midpoint value of the interval is used as a proxy for s.
A second problem is related to the endogeneity of explanatory variables. As explained
by Harrison et al. (2008), endogeneity is derived from nominal sales growth rather than real
output growth, as suggested by the theoretical model. However, in the case where it was
possible to access real output growth in old and new products, there is another endogeneity
problem. Innovations are the result of investment decisions (such as R&D), which have to
be decided by firms in advance. These decisions depend on the firm’s productivity, which
can be thought of as an unobservable of two components: firm’s attributes that are mainly
constant over time (such as managerial skills or organizational capital) and productivity
shocks (which might lead to the firm to reduce labor costs). Given that innovation
investments are correlated with firm productivity, then innovation outputs will be also
correlated with firm productivity. As firm productivity also shows up in the error term of
the labor demand equation, innovation outputs will be endogenous, leading to an
identification problem.
In a dynamic setting, lags of the explanatory variables can be used as instruments.
Given that Chilean panel data information is available only for a reduced number of firms
through all the waves of the innovation surveys, we use an identification strategy based on
external instruments. To be valid instruments, these exogenous variables must be highly
correlated with innovation decisions but not with employment growth directly. There are
several potential instruments that have been considered in the literature: (i) the importance
of external sources as the origin of ideas for innovation, (ii) the importance of regulations
17
as objectives for innovation decisions, and (iii) the relevance of obstacles to innovation. All
of these variables are assumed to be exogenous and correlated with innovation variables.
The exogeneity of obstacles to innovation based on the firm’s perception, however, has
been challenged because the more innovative firms may be more aware of these obstacles
(Savignac, 2008; Mohnen et al., 2008; Iammarino, et al. 2007; Galia and Legros, 2004).
This literature then casts doubts on the validity of these variables as instruments for
innovation activities. Because of this concern, we use as instruments the average of these
variables for firms located in the same region. The variables at the plant level are defined as
the average of the importance for three types of obstacles: economic (technological risk,
very long-term return, etc.), workforce-related (low skills, lack of experience, etc.), and
other factors (absence of technological or market-related information, innovation is easily
imitated, etc.).7
The identification strategy used in this paper captures the idea that there are common
exogenous factors for firms in a particular region, and it would be less likely to be affected
by reverse causality. The assumption herein is that there is a high correlation between
obstacles perceived by a firm and others in the same region because they face common and
exogenous barriers to innovation. These can be associated with regional differences in
credit access, supply of skilled labor and researchers at universities and technological
centers, and public funds allocated for innovation.
Benavente and Lauterbach (2008) use as an instrument for innovation the degree of
firms’ utilization of novel inputs as an origin of innovation ideas. However, these variables
are defined only for innovative firms, reducing the number of available observations and
generating sample selection problem because the estimation can only be performed for the
group of innovative firms. Chilean innovation surveys ask about innovation sources and
objectives of the innovation for the group of firms declaring that they are undertaking some
type of innovation.
7
The perception of each obstacle is defined on a five-point scale from 0 (not important) to 4 (highly
important). In a previous version of this paper, we used alternatively the average of these three obstacles
across firms in the same sector and the average across firms of similar size. The results are similar and
available upon request.
18
Table 6. Naive Regression on the Effect of Innovation on Employment Quantity
Dependent variable: l (Employment growth-yearly)
Sector
All
Manuf
All
Manuf
All
Manuf
All
Manuf
Small
Manuf
Small
Manuf
Small
Manuf
Small
Manuf
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
2.527***
2.788***
3.463***
2.839***
(0.958)
(0.943)
(0.9189
(0.9379)
Regression
Constant
2.109*** 2.621*** 3.761*** 2.566***
(0.642)
(0.65)
(0.601)
(0.603)
TPP (product or process innovator) 3.144***
4.268***
(0.765)
Product innovator
Process innovator only
(1.242)
3.664***
4.206***
(0.829)
(1.417)
-0.426
-2.319**
0.367
-1.001
(1.078)
(0.981)
(2.113)
(2.030)
TPP red (product and process
innovator)
Product innovator (only)
Real sales growth (g1-Π)
3.736***
4.690***
(0.741)
(1.397)
-0.254
-2.224
(1.845)
(3.207)
0.181*** 0.188*** 0.176*** 0.187***
0.201***
0.203***
0.191***
0.204***
(0.021)
(0.021)
(0.02)
(0.021)
(0.035)
(0.036)
(0.035)
(0.035)
Time dummies
yes
yes
yes
Yes
yes
yes
yes
yes
2-digit industry dummies
yes
yes
yes
Yes
yes
yes
yes
yes
R-squared
0.1
0.104
0.096
0.104
0.118
0.115
0.105
0.118
2049
2049
2049
2049
652
652
652
652
Number of firms
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent.
4.2 Econometric Results
Table 6 presents the results for specifications in which, in addition to dummy variables for
innovation, a measure of real sales growth of old products is included. The results are
similar for both samples. First, the results show that old product sales influence
employment growth positively and significantly. Second, they show that the effect of joint
process and product innovation is positive and significant, but it seems to be mostly
explained by product innovation. Note that most of the regressions show that process
innovation by itself does not seem to affect employment growth.
19
Table 7 presents the results for similar specifications to those in Harrison et al. (2008)
for all firms and for small firms only, as well as results dividing the sample between highand low-tech industries.8 In all of these regressions, the dependent variable is employment
growth minus the real growth of sales due to the unchanged products. The first column
includes as explanatory variables a dummy for process innovation and real sales growth
due to new products only. In the subsequent columns, we control for foreign ownership,
localization, and changes in wages.
We find that the effect of process innovation is negative and significant for all firms,
but not for small firms or for firms in low-tech-industries. We also find that sales growth
due to product innovation is positively associated with employment growth for all samples
under analysis, and that coefficients are very similar across all samples. Then, the results
show that there are no relevant differences in this relationship across groups of firms.
Finally, these estimations show that foreign ownership and localization are not significant
explanatory variables for employment growth, and that wage increases, as expected,
generate a displacement effect on employment.
The results of IV estimations are presented in Table 8. As previously explained, the
instruments for the sales growth of new products are the importance of three obstacles for
innovation averaged across firms located in the same region. These results confirm the
existence of a positive and significant effect of the sale of new products on employment
growth across all samples analyzed. Note that the coefficient for this variable is much
larger than the one in OLS estimations. This finding is consistent with the correction of the
downward bias in OLS related to endogeneity and price mismeasurement (Harrison et al.,
2008). Also, these IV results show that process innovation does not seem to matter for
employment growth in all of the specifications.
8
We divide the 2-digit industries into two groups by calculating the Innovation Expenditures over turnover
for the entire sample. Those sectors below or in the median are low tech, while the rest is classified as high
tech.
20
Table 7. Effect of Innovation on Employment Quantity
Dependent variable: l -(g1-Π)
Sector
Regression
Constant
Process innovation only (d)
Sales growth due to new products
(g2)
All
Manuf
All
Manuf
All
Manuf
High
Tech
High
Tech
High
Tech
Low
Tech
Low
Tech
Low
Tech
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
1.855***
1.997**
3.251***
2.268**
3.136**
5.149***
1.311
1.131
2.755***
2.916**
4.113***
4.591***
(0.698)
(0.825)
(0.789)
(1.039)
(1.326)
(1.253)
(0.824)
(0.977)
(0.929)
(1.295)
(1.555)
(1.499)
-2.755**
-2.780**
-2.566**
-3.373
-3.346
-3.406
-2.976**
-2.943**
-2.965**
-2.25
-2.457
-1.858
(1.274)
(1.275)
(1.181)
(2.716)
(2.717)
(2.414)
(1.50)
(1.492)
(1.352)
(2.362)
(2.406)
(2.309)
0.833***
0.833***
0.839***
0.714***
0.706***
0.759***
0.813***
0.813***
0.817***
0.883***
0.879***
0.888***
(0.034)
(0.034)
(0.031)
(0.083)
(0.084)
(0.074)
(0.039)
)0.039)
(0.036)
(0.064)
(0.064)
(0.061)
-0.13
-0.409
3.064
2.724
0.061
-0.395
-0.897
-0.721
(1.231)
(1.191)
(4.309)
(4.323)
(1.347)
(1.269)
(2.874)
(2.977)
-0.275
0.083
-2.318
-2.048
0.395
0.613
-2.208
-1.443
(0.889)
(0.849)
(1.718)
(1.582)
(1.026)
(0.979)
(1.775)
(1.693)
Foreign owned (10% or more)
Located in the capital (capreg)
Growth wage bill per worker
Small Manuf Small Manuf Small Manuf
-0.333***
-0.448***
-0.347***
-0.292***
(0.035)
(0.065)
(0.039)
(0.077)
2-digit industry dummies
yes
yes
yes
yes
yes
Yes
yes
yes
yes
yes
yes
yes
Time dummies
yes
yes
yes
yes
yes
Yes
yes
yes
yes
yes
yes
yes
R-squared
0.27
0.27
0.331
0.159
0.163
0.254
0.268
0.268
0.339
0.279
0.281
0.318
Number of firms
2049
2049
2049
652
652
652
1417
1417
1417
632
632
632
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent.
Table 8. Effect of Innovation on Employment Quantity
Dependent variable: l -(g1-Π)
List of instrument(s) used: Innovation obstacles (3) averaged by region
Sector
Regression
Constant
Process innovation only (d)
Sales growth due to new products (g2)
All Manuf
All Manuf
All Manuf
Small
Manuf
Small
Manuf
Small
Manuf
IV
IV
IV
IV
IV
IV
-1.989
-2.016
-0.826
-0.61
-2.125
1.237
-2.754
-2.733
-0.332
1.049
1.697
1.608
(2.794)
(3.0)
(2.909)
(2.711)
(4.701)
(3.843)
(3.507)
(3.368)
(3.056)
(3.189)
(4.206)
(4.234)
High Tech High Tech
IV
IV
High Tech
IV
Low Tech Low Tech
IV
IV
Low Tech
IV
0.297
0.333
0.633
-3.417
-3.38
-3.443
0.028
-0.076
-0.657
-0.551
-0.517
0.581
(2.495)
(2.572)
(2.517)
(2.698)
(2.921)
(2.621)
(2.951)
(2.835)
(2.589)
(3.398)
(3.787)
(3.86)
1.740***
1.751***
1.780***
1.813*
2.141*
1.918*
1.734**
1.695**
1.528**
1.356*
1.403*
1.540*
(0.633)
(0.653)
(0.638)
(0.927)
(1.205)
(1.052)
(0.765)
(0.728)
(0.664)
(0.744)
(0.846)
(0.858)
Foreign owned (10% or more)
Located in the capital (capreg)
-0.36
-0.653
5.208
4.393
0.362
-0.16
-1.984
-2.066
(1.449)
(1.416)
(4.699)
(4.179)
(1.620)
(1.483)
(3.332)
(3.353)
0.048
0.425
0.865
0.546
0.237
0.489
-1.399
-0.402
(1.040)
(1.017)
(3.394)
(3.012)
(1.174)
(1.075)
(2.231)
(2.281)
Growth wage bill per worker
-0.343***
-0.517***
-0.352***
-0.305***
(0.03)
(0.087)
(0.032)
(0.058)
2-digit industry dummies
yes
yes
Yes
yes
Yes
yes
yes
Yes
yes
yes
yes
Time dummies
yes
Yes
Yes
yes
Yes
yes
yes
Yes
yes
yes
yes
yes
0.247
0.247
0.294
0.1401
0.1392
0.2104
0.2458
0.2464
0.3078
0.256
0.2568
0.29
R-squared
Number of firms
yes
2049
2049
2049
652
652
652
1417
1417
1417
632
632
632
F test, g2
46.210
35.910
32.350
8.100
6.940
6.750
30.250
23.570
21.220
16.530
13.140
11.840
p-value
Davidson-MacKinnon test of
exogeneity
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
2.799
2.709
3.100
1.889
2.235
1.724
2.002
1.982
1.434
0.446
0.429
0.688
P-Value
0.095
0.100
0.079
0.170
0.135
0.190
0.157
0.159
0.231
0.504
0.513
0.407
Sargan (m) 1
2.315
2.549
2.018
0.870
0.738
0.690
0.516
0.490
0.395
3.443
4.442
4.355
Prob. Value
0.314
0.280
0.365
0.647
0.692
0.708
0.773
0.782
0.821
0.179
0.109
0.113
Diff. Sargan (m) 1
2.803
2.716
3.109
1.901
2.255
1.744
2.008
1.991
1.442
0.450
0.435
0.698
Prob. Value
0.094
0.099
0.078
0.168
0.133
0.187
0.157
0.158
0.230
0.502
0.510
0.404
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent.
22
Table 9 shows the contribution of innovation to employment growth. In contrast to the
findings of Harrison et al. (2008) for European countries, in the Chilean case the net
contribution of net product innovation is negative in the case of IV and for the entire
sample and also for high- and low-tech industries. Only for small firms do we find a
positive contribution of net product innovation. For all samples and IV estimations, the
decomposition shows that the contribution of the productivity trend in the production of old
products is positive. The results, shown in Table 10, are very similar when this
decomposition is carried out for the recovery phase of the cycle.9
Table 9. Employment Growth Decomposition
Contributions of innovation to employment growth.
Manufacturing. Period 1993–2006.
All Manufacturing
Small firms in High-tech firms in Low-tech firms in
Manufacturing Manufacturing
Manufacturing
OLS
IV
OLS
IV
OLS
IV
OLS
IV
Firms employment growth
-0.181
-0.181
1.447 1.447
-0.127
-0.127
-0.302
-0.302
Productivity trend in production of old products
-0.161
0.904
1.788 0.814
-0.333
0.019
0.219
2.759
Gross effect of process innovation in production of old products
-0.105
-0.132 -0.136 -0.136 -0.105
-0.113
-0.098
-0.183
Output growth of old products contribution
-0.039
-0.039 -0.029 -0.029
0.109
0.109
-0.373
-0.373
Net contribution of product innovation
Contribution of old products by product innovators
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
0.124
-6.044
6.167
-0.914 -0.177 0.798 0.201
-6.044 -3.147 -3.147 -5.980
5.129 2.970 3.945 6.181
-0.144
-5.980
5.836
-0.051
-6.187
6.136
-2.505
-6.187
3.682
Table 10. Employment Growth Decomposition for the Recovery Phase of the Cycle
Contributions of innovation to employment growth.
Manufacturing. Period 2001-2006.
Firms employment growth
All
Manufacturing
OLS
IV
0.293 0.293
Small firms in High-tech firms Low-tech firms
Manufacturing in Manufacturing in Manufacturing
OLS
IV
OLS
IV
OLS
IV
1.801 1.801 -0.019 -0.019 0.980 0.980
Productivity trend in production of old products
0.219
1.302
1.180
-0.266
1.284
5.463
Gross effect of process innovation in production of old products
-0.177 -0.321 -0.173
-0.173
-0.157 -0.268 -0.209
-0.439
Output growth of old products contribution
0.407
0.581
0.581
0.619
-0.061
-0.061
Net contribution of product innovation
Contribution of old products by product innovators
-0.156 -4.387 0.092
-3.633 -3.633 -2.058
0.213
-2.058
-0.215 -4.663 -0.034
-3.817 -3.817 -3.228
-3.983
-3.228
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
3.477
2.272
3.602
-0.754
4.594
0.407
-0.754
2.150
4.294
0.619
-0.846
3.195
Tables 11 and 12 present some robustness checks. Specifically, we study the
exogeneity of d and we test for a structural change in the slope of product innovations.
According to the Davidson-MacKinnon test, we reject the exogeneity of process innovation
only (d), and we confirm that sales growth due to new products (g2) affects total
9
This corresponds to the period 2001–2006.
employment positively and significantly (Table 11). In contrast, when the interaction terms
are included, most of the parameters are not statistically significant (Table 12). Then, our
results reject the idea that there can be differential effects of product innovation depending
on whether firms are conducting process innovations.
Table 11. Testing the Endogeneity of d, Dependent Variable: l -(g1-Π)
Sector
All Manuf. All Manuf. All Manuf.
Type of Labor
All
Skilled
Unskilled
Regression
IV
IV
IV
0.377
-6.378
5.873
(5.647)
(6.786)
(6.452)
-4.423
14.931
-29.793
(20.470)
(28.568)
(27.162)
1.775***
1.702*
1.540
(0.640)
(1.029)
(0.978)
-0.697
2.613
-0.982
(1.430)
(2.988)
(2.841)
-0.031
-0.061**
(0.033)
(0.031)
-3.000
0.406
(4.037)
(3.839)
Yes
yes
Constant
Process innovation only (d)
Sales growth due to new products (g2)
Foreign owned (10% or more)
Located in the capital (capreg)
0.166
(1.456)
Growth wage bill per worker
-0.341***
(0.031)
Capital stock growth
Fully foreign owned
2-digit industry dummies
yes
Time dummies
yes
Yes
yes
R-sSquared
0.295
0.1219
0.125
Number of firms
2049
1973
1973
F test, g2
33.260
32.120
32.120
p-value
0.000
0.000
0.000
F test, d
7.860
6.460
6.460
p-value
0.000
0.000
0.000
Davidson-MacKinnon test of exogeneity
1.593
0.666
0.867
P-Value
0.204
0.514
0.421
Sargan (m) 1
1.947
0.490
3.421
Prob. Value
0.163
0.484
0.064
Diff. Sargan for d (m) 1
0.062
0.198
1.331
Prob. Value
0.803
0.656
0.249
Notes: Robust standard errors in parentheses, Significance level: *** 1 percent, ** 5 percent, and * 10
percent. List of instrument(s) used: Innovation obstacles (3) averaged by region
24
Table 12. Testing for a Structural Change, Dependent Variable: l -(g1-Π)
Sector
All Manuf. All Manuf. All Manuf.
Type of Labor
All
Skilled
Unskilled
Regression
IV
IV
IV
4.021***
0.586
3.722***
(1.186)
(1.406)
(1.319)
Constant
Process innovation only (d)
Sales growth due to new products (g2)
g2 x product & process innovation
Foreign owned (10% or more)
Located in the capital (capreg)
0.501
-0.12
1.337
(3.792)
(5.066)
(4.752)
-0.171
0.427
-0.700
(1.143)
(1.436)
(1.347)
0.914
0.298
1.352
(1.052)
(1.314)
(1.232)
-0.637
2.769
-0.498
(1.267)
(2.691)
(2.524)
-0.462
(1.07)
Growth wage bill per worker
-0.326***
(0.027)
Capital stock growth
Fully foreign owned
-0.009
-0.052**
(0.023)
(0.022)
-3.331
-0.695
(3.564)
(3.344)
2-digit industry dummies
Yes
yes
yes
Time dummies
Yes
yes
yes
0.2497
0.1268
0.0801
2049
1973
1973
F test, g2
55.280
52.030
52.030
p-value
0.000
0.000
0.000
F test, g2 x product & process innovation
79.510
75.990
75.990
p-value
0.000
0.000
0.000
Davidson-MacKinnon test of exogeneity
0.810
0.125
0.799
P-Value
0.445
0.883
0.450
Sargan (m) 1
6.723
2.165
4.988
Prob. Value
0.151
0.705
0.289
Diff. Sargan (m) 1
1.627
0.251
1.605
Prob. Value
0.443
0.882
0.448
R-Squared
Number of firms
Robust standard errors in parentheses, Significance level: *** 1 percent, ** 5 percent, and * 10 percent. List of instrument(s) used:
Innovation obstacles (3) averaged by region.
25
5. Innovation and Employment Quality
The section analyzes the impact of innovation on the composition of the workforce.
Depending on the skill bias of innovation, its impact may be different for skilled and
unskilled workers. Whether innovation is skilled-biased, as several empirical and
theoretical studies argue (see, for example, Card and Dinardo, 2002; Acemoglu, 1998),10
higher innovation could be associated with lower employment growth for unskilled workers
and higher employment growth for skilled workers.
For analyzing the effect of innovation on employment composition, we estimate the
following equation:
l qj % ( g 1 % $ ) = # 0qj + # 1qj d + " 1qj g 2 + ! j
The dependent variable is employment growth (minus old product real sales growth)
for both types of workers: skilled and unskilled. In the basic equation, the explanatory
variables are a dummy variable for process innovation (d), real sales growth of new
products (g2), capital stock growth, a dummy variable for foreign-owned firms, a dummy
variable for full foreign-owned firms, and survey year and 2-digit industries fixed effects.
Figure 1
Panel B: Skilled Workers on Total Employment
.3
.2
.35
.25
.4
.3
.45
.35
.4
.5
.45
.55
Panel A: Skilled Wage Bill on Total Labor Costs.
1995
2000
year
Simple Average
1995
2005
2000
year
Simple Average
Median
2005
Median
Source: Authors’ elaboration based on Annual Survey of Manufactures (ENIA).
The Chilean innovation surveys lack information the skill level of workers. However,
each of these surveys is matched with the Annual Survey of Manufacturers. Specifically,
we use the broad classification of skilled and unskilled workers as white-collar and bluecollar workers directly from the Annual Survey of Manufacturers. To indicate the evolution
10
For the case of Chile, see Gallego (2011).
26
of skill composition in Chilean industry, we show the evolution of the skilled wage bill as a
proportion of total labor costs (Figure 1A) and skilled workers as a proportion of total
employment (Figure 1B) for the period 1994–2005. In both cases, the evidence suggests a
process of skill upgrading in Chilean manufacturing industries.
Table 13. Descriptive Statistics for Employment Composition, All Manufacturing
1993–2006
Mean
Median
Standard
deviation
Minimum
Maximum
All firms
0.35
0.29
0.22
0.02
0.99
Noninnovators (no process or product innovations)
0.36
0.29
0.24
0.02
0.98
Process only innovators (non-product innovators)
0.35
0.28
0.25
0.04
0.98
Organizational change innovator (non-product innovators)
0.35
0.29
0.24
0.02
0.96
Product innovators
0.34
0.29
0.21
0.02
0.99
All firms
-0.18
0.00
14.65
-67.43
62.14
Noninnovators (no process or product innovations)
0.80
0.00
14.59
-66.79
58.32
Process only innovators (non-product innovators)
2.12
3.62
14.02
-41.01
34.01
Organizational change innovator (non-product innovators)
-1.96
-0.65
17.79
-53.39
62.14
Product innovators
-0.52
-0.39
13.89
-67.43
58.59
All firms
0.27
0.00
25.05
-110.29
97.30
Noninnovators (no process or product innovations)
0.48
0.00
24.97
-107.00
95.85
Process only innovators (non-product innovators)
2.62
0.00
20.71
-70.99
86.14
Organizational change innovator (non-product innovators)
-0.29
0.00
27.56
-102.71
97.30
Product innovators
0.09
0.00
24.82
-110.29
95.38
All firms
-0.62
0.00
23.41
-104.17
107.97
Noninnovators (no process or product innovations)
0.94
0.00
23.42
-97.30
103.97
Process only innovators (non-product innovators)
1.11
2.97
20.12
-77.53
53.75
Organizational change innovator (non-product innovators)
-2.72
-0.28
27.71
-104.17
98.31
Product innovators
Sources: Authors’ elaboration based on Innovation Surveys.
-1.19
-0.24
22.54
-100.41
107.97
Share of skilled labor (at the beginning of each survey)
Employment (total) growth (%)
Skilled labor growth (%)
Unskilled labor growth (%)
Tables 13 and 14 present descriptive statistics for the entire sample and small firms.
For all firms, the average and median share of skilled labor is 35 percent and 29 percent,
respectively, and very similar across types of innovators. These figures are also similar to
those in the sample of small firms, with an average and median of 36 percent and 29
percent, respectively. For the entire sample, the average increase in skilled labor was 0.3
percent and -0.6 percent for unskilled labor.
27
There are important differences across firms. The highest growth in skilled and
unskilled labor was experienced by process only innovators (non-product innovators) and
the lowest for organizational change innovators (non-product innovators). In the case of
small firms, the average growth in unskilled labor (2.3 percent) was higher than the average
growth in skilled labor (0.1 percent), and similar to the entire sample. The highest
expansion in both types of workers corresponded to process only innovators (non-product
innovators).
Table 14. Descriptive Statistics for Employment Composition: All Manufacturing
1993–2006
Mean
Median
Standard
deviation
Minimum
Maximum
All firms
0.36
0.30
0.23
0.03
0.98
Noninnovators (no process or product innovations)
0.38
0.31
0.24
0.03
0.98
Process only innovators (non-product innovators)
0.34
0.26
0.26
0.04
0.94
Organizational change innovator (non-product innovators)
0.35
0.29
0.23
0.06
0.96
Product innovators
0.34
0.27
0.21
0.05
0.96
All firms
1.45
0.00
15.22
-53.39
58.32
Noninnovators (no process or product innovations)
1.49
0.00
14.62
-49.04
58.32
Process only innovators (non-product innovators)
4.54
3.79
11.05
-13.88
34.01
Organizational change innovator (non-product innovators)
-1.08
0.00
19.98
-53.39
46.12
Product innovators
1.83
0.00
14.96
-44.12
52.23
All firms
0.14
0.00
25.01
-102.71
97.30
Noninnovators (no process or product innovations)
-0.24
0.00
23.86
-93.59
95.85
Process only innovators (non-product innovators)
10.19
0.00
24.17
-20.27
86.14
Organizational change innovator (non-product innovators)
-2.02
0.00
30.51
-102.71
97.30
Product innovators
0.38
0.00
25.15
-85.24
78.43
All firms
2.27
0.00
23.81
-97.30
107.97
Noninnovators (no process or product innovations)
2.26
0.00
23.87
-97.30
103.97
Process only innovators (non-product innovators)
3.34
1.85
15.11
-21.14
40.55
Organizational change innovator (non-product innovators)
-1.91
0.00
25.85
-73.32
53.04
Product innovators
Sources: Authors’ elaboration based on Innovation Surveys.
3.61
0.00
23.87
-69.31
107.97
Share of skilled labor (at the beginning of each survey)
Employment (total) growth (%)
Skilled labor growth (%)
Unskilled labor growth (%)
The concerns about endogeneity analyzed in the employment equation are also relevant
for this employment quality regression. The identification strategy used is the same as
before. In the absence of panel data, we use external instrumental variables. These are the
28
ones explained in previous section under the assumption that there are exogenous variables
explaining innovation decisions that are not correlated with the error term of the equation.
These are the average region-specific variables related to three types of obstacles for
innovation.
Table 15. Effect of Innovation on Employment Quality (OLS Approach)
All
Skilled
OLS
All
Unskilled
OLS
Small
Skilled
OLS
Small
Unskilled
OLS
H-T
Skilled
OLS
H-T
Unskilled
OLS
L-T
Skilled
OLS
L-T
Unskilled
OLS
Constant
0.045
(1.083)
2.823***
(1.020)
-0.079
(1.523)
3.163**
(1.453)
-0.478
(1.269)
3.106**
(1.280)
1.350
(2.051)
2.219
(1.697)
Process innovation only (d)
-0.894
(1.837)
-3.388*
(1.785)
0.52
(3.693)
-5.836*
(3.328)
0.702
(2.190)
-5.015**
(2.218)
-4.558
(3.254)
-0.157
(2.977)
0.841***
(0.052)
0.814***
(0.048)
0.793***
(0.104)
0.695***
(0.109)
0.872***
(0.061)
0.782***
(0.06)
0.783***
(0.101)
0.885***
(0.077)
-0.01
(0.029)
-0.050*
(0.026)
-0.063
(0.05)
-0.046
(0.034)
-0.002
(0.037)
-0.045
(0.034)
-0.026
(0.042)
-0.059
(0.039)
Sector
Regression
Sales growth due to new
products (g2)
Capital stock growth
Foreign owned (10% or more)
Fully foreign owned
2-digit industry dummies
Time dummies
R-squared
2.854
0.113
4.147
0.972
2.571
-0.72
3.794
1.838
(2.769)
(2.606)
(8.723)
(6.402)
(3.304)
(3.188)
(5.142)
(4.563)
-3.342
(3.478)
-1.053
(3.448)
4.908
(10.788)
8.828
(11.436)
-1.587
(4.026)
0.264
(4.074)
-13.448**
(6.750)
-5.409
(6.128)
Yes
Yes
0.144
yes
yes
0.163
yes
yes
0.116
yes
yes
0.112
yes
yes
0.143
yes
yes
0.151
Yes
Yes
0.159
yes
yes
0.198
Number of firms
1973
1973
625
625
1363
1363
610
610
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent. (3) H-T and L-T
means High and Low-Technology Industries respectively.
The OLS estimations for all and small firms, and high-tech and low-tech industries are
presented in Table 15. We find that process innovation has a negative and significant effect
on unskilled employment growth except in low-tech industries. In contrast, our findings
show that product innovation—measured by sales growth due to new product—increased
both types of employment, and the parameter is practically the same for skilled and
unskilled workers. This is valid across all the samples of firms. The IV estimations are
shown in Table 16 and they are not satisfactory. Most of the coefficients turn out to be not
significant.
Tables 17 through 20 present the decomposition of the effect of innovation-related
variables on employment growth for both types of workers, across types of firms, and
across industries. Given the unsatisfactory results for IV estimations, we discuss the results
from OLS regressions. These results show that for skilled workers, the net contribution of
product innovation is positive and large for all firms and those in high-tech industries. By
29
contrast, the contribution for small firms, although positive, is of low magnitude, and it is
negative for firms in low-tech industries. The contribution of process innovation in old
products to employment growth is generally negative, with the exception of firms in hightech industries.
Table 16. Effect of Innovation on Employment Quality (IV Approach).
All
Skilled
IV
-4.072
(4.414)
All
Unskilled
IV
0.763
(3.778)
Small
Skilled
IV
-1.299
(3.555)
Small
Unskilled
IV
1.025
(3.307)
H-T
Skilled
IV
-5.059
(5.654)
H-T
Unskilled
IV
3.488
(4.794)
L-T
Skilled
IV
-1.782
(5.10)
L-T
Unskilled
IV
-0.923
(4.441)
2.296
(3.804)
-1.792
(3.255)
0.493
(3.442)
-5.885*
(3.202)
3.983
(4.551)
-5.288
(3.859)
-1.776
(5.306)
2.635
(4.620)
1.810*
(1.003)
1.299
(0.858)
1.237
(1.160)
1.472
(1.080)
1.906
(1.236)
0.696
(1.048)
1.581
(1.196)
1.686
(1.041)
Capital stock growth
-0.031
(0.033)
-0.061**
(0.028)
-0.064*
(0.035)
-0.048
(0.033)
-0.027
(0.043)
-0.043
(0.036)
-0.04
(0.045)
-0.073*
(0.039)
Foreign owned (10% or more)
2.254
(2.888)
-0.187
(2.472)
4.845
(6.663)
2.194
(6.199)
2.424
(3.364)
-0.708
(2.852)
2.533
(5.525)
0.574
(4.810)
Fully foreign owned
-2.526
(3.909)
-0.645
(3.345)
5.096
(10.418)
9.158
(9.692)
-0.655
(4.471)
0.186
(3.791)
-14.406
(8.792)
-6.37
(7.655)
2-digit industry dummies
Time dummies
R-Squared
Number of firms
Yes
Yes
0.130
1973
yes
yes
0.149
1973
yes
yes
0.097
625
yes
yes
0.098
625
yes
yes
0.134
1363
yes
yes
0.145
1363
Yes
Yes
0.135
610
yes
yes
0.170
610
F test, g2
p-value
Davidson-MacKinnon test of
exogeneity
p-Value
31.130
0.000
31.130
0.000
5.940
0.000
5.940
0.000
20.400
0.000
20.400
0.000
11.150
0.000
11.150
0.000
1.101
0.294
0.337
0.562
0.151
0.698
0.577
0.448
0.844
0.359
0.007
0.934
0.500
0.480
0.691
0.406
Sargan (m) 1
Prob. Value
0.685
0.710
5.523
0.063
1.466
0.481
0.559
0.756
0.596
0.742
1.772
0.412
0.574
0.750
6.478
0.039
Sector
Regression
Constant
Process innovation only (d)
Sales growth due to new products
(g2)
Diff. Sargan (m) 1
1.105
0.339
0.153
0.585
0.849
0.007
0.507
0.701
Prob. Value
0.293
0.561
0.695
0.445
0.357
0.934
0.476
0.403
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent, (3) H-T and L-T
means High and Low-Technology Industries respectively.
For unskilled workers, we find that the net contribution of product innovation is
generally negative. Thus, these results are consistent with the idea that product innovation
is skill-biased in the Chilean case. These differences in the contribution of product
innovation to employment growth across types of workers are valid for all the samples and
for the entire period under analysis, but they do not hold for the shorter period 2001–2006
(Tables 19 and 20).
30
Table 17. Employment Growth Decomposition for Skilled Labor
Contributions of innovation to employment growth.
Manufacturing. Period 1993–2006.
Firms employment growth
All
Manufacturing
OLS
IV
0.267
0.267
Small firms in
Manufacturing
OLS
IV
0.142
0.142
High-tech firms Low-tech firms in
in Manufacturing Manufacturing
OLS
IV
OLS
IV
0.175
0.175
0.472
0.472
Productivity trend in production of old products
Gross effect of process innovation in production of old
products
0.157
-0.411
0.185
-0.434
-0.410
-0.676
1.451
0.456
-0.058
-0.044
-0.030
-0.030
0.022
0.028
-0.282
-0.249
Output growth of old products contribution
-0.039
-0.039
-0.029
-0.029
0.109
0.109
-0.373
-0.373
Net contribution of product innovation
Contribution of old products by product innovators
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
0.207
-6.044
6.251
0.760
-6.044
6.804
0.016
-3.147
3.163
0.635
-3.147
3.782
0.454
-5.980
6.433
0.714
-5.980
6.694
-0.324
-6.187
5.863
0.637
-6.187
6.824
Table 18. Employment Growth Decomposition for Unskilled Labor
Contributions of innovation to employment growth.
Manufacturing. Period 1993–2006.
All
Manufacturing
OLS
IV
-0.624 -0.624
Small firms in
Manufacturing
OLS
IV
2.275
2.275
Productivity trend in production of old products
Gross effect of process innovation in production of old
products
-0.260
1.560
2.735
1.085
-0.184
1.547
-0.451
1.775
-0.130
-0.177
-0.210
-0.211
-0.174
-0.213
0.002
-0.072
Output growth of old products contribution
-0.039
-0.039
-0.029
-0.029
0.109
0.109
-0.373
-0.373
Net contribution of product innovation
Contribution of old products by product innovators
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
-0.194
-6.044
5.849
-1.967
-6.044
4.076
-0.222
-3.147
2.925
1.429
-3.147
4.576
-0.259
-5.980
5.721
-1.951
-5.980
4.028
-0.065
-6.187
6.122
-2.217
-6.187
3.970
Firms employment growth
High-tech firms in Low-tech firms in
Manufacturing
Manufacturing
OLS
IV
OLS
IV
-0.507 -0.507 -0.887 -0.887
Table 19. Employment Growth Decomposition for Skilled Labor for Recovering
Phase of Cycle
Contributions of innovation to employment growth.
Manufacturing. Period 2001–2006.
Firms employment growth
OLS
0.585
IV
0.585
Small firms in
Manufacturing
OLS
IV
0.606
0.606
All Manufacturing
High-tech firms in Low-tech firms in
Manufacturing
Manufacturing
OLS
IV
OLS
IV
0.590
0.590
0.574
0.574
Productivity trend in production of old products
Gross effect of process innovation in production of old
products
0.492
1.777
-0.070
-0.623
-0.028
0.802
1.860
2.443
-0.176
-0.218
-0.028
-0.029
-0.044
-0.064
-0.611
-0.643
Output growth of old products contribution
0.407
0.407
0.581
0.581
0.619
0.619
-0.061
-0.061
Net contribution of product innovation
Contribution of old products by product innovators
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
-0.138
-3.633
3.495
-1.381
-3.633
2.252
0.123
-2.058
2.182
0.678
-2.058
2.736
0.042
-3.817
3.859
-0.767
-3.817
3.050
-0.614
-3.228
2.614
-1.165
-3.228
2.064
31
Table 20. Employment Growth Decomposition for Unskilled Labor for Recovering
Phase of Cycle
Contributions of innovation to employment growth.
Manufacturing. Period 2001–2006.
All
Manufacturing
OLS
IV
-0.277 -0.277
Small firms in
Manufacturing
OLS
IV
2.423
2.423
Productivity trend in production of old products
Gross effect of process innovation in production of old
products
-0.350
4.025
1.923
1.802
-0.709
3.851
0.434
4.614
-0.177
-0.321
-0.173
-0.173
-0.157
-0.268
-0.209
-0.439
Output growth of old products contribution
0.407
0.407
0.581
0.581
0.619
0.619
-0.061
-0.061
Net contribution of product innovation
Contribution of old products by product innovators
Contribution of new products by product innovators
Source: Authors’ elaboration based on Innovation Surveys.
-0.156
-3.633
3.477
-4.387
-3.633
-0.754
0.092
-2.058
2.150
0.213
-2.058
2.272
-0.215
-3.817
3.602
-4.663
-3.817
-0.846
-0.034
-3.228
3.195
-3.983
-3.228
-0.754
Firms employment growth
High-tech firms in Low-tech firms in
Manufacturing
Manufacturing
OLS
IV
OLS
IV
-0.462 -0.462
0.131
0.131
6. Innovation Strategies and Employment
In this section, we analyze the effect of innovation strategies on employment. To do that,
we use questions from the Chilean Innovation Survey that looked at the types of innovation
strategies conducted by the firms.11 Specifically, we define two dummy variables for
internal (make) and external (buy) innovation, using the response to the questions asking
whether during the last two years the firm undertook the following innovative activities:
1. Carried out R&D within the firm
2. Carried out R&D outside the firm
3. Purchased external knowledge (patents, licenses, know-how)
“Make only” is a dummy variable taking the value 1 if the response to question 1 is
affirmative and responses to questions 2 and 3 are negative; and 0 otherwise. “Buy only” is
a dummy variable taking the value 1 if the response to questions 2 and 3 is affirmative and
response to question 1 is negative; and 0 otherwise. Finally, to look at potential reinforcing
effects of both activities, we also included a dummy variable for firms claiming to carry out
both make and buy innovations. In such a case, make and buy is 1 if response to question 1
is affirmative and any of responses to other questions is positive; otherwise, it is 0.
11
Unfortunately, given the requirements for defining these variables, we have only one available survey
where this question was introduced. Therefore, in this section, we report only the analysis for the 2007
innovation survey.
32
Table 21. Descriptive Statistics for Strategies. All Manufacturing
2004–2006
Share of firms pursuing each type of strategy by type of firm
(%)
All firms
Buy
Only
4.43
Make
Only
12.81
Make
and Buy
9.22
Noninnovators (no process or product innovations)
1.07
2.78
1.07
Process only innovators (non-product innovators)
7.41
12.96
12.96
Organizational change innovator (non-product innovators)
7.14
13.49
11.90
Product innovators
10.11
37.23
26.60
Source: Authors’ elaboration based on Innovation Surveys.
Tables 21 and 22 present descriptive statistics for the entire sample and for small firms,
respectively, and for the three different innovation strategies: buy only, make only, and
make and buy. For both samples, the most prevalent strategy is “make only”, representing
12.8 percent for all firms and 7.1 percent in small firms. This is valid across the different
types of firms. Notably, product innovators are more likely to use the three strategies
compared with the rest of the firms. In both samples, about 30 percent of product
innovators follow a make only strategy. The prevalence of these strategies, as expected, is
very low for noninnovators (no process or product innovations).
Table 22. Descriptive Statistics for Strategies. Small Firms in Manufacturing
2004–2006
Share of firms pursuing each type of strategy by type of
firm (%)
All firms
Noninnovators (no process or product innovations)
Buy
Only
1.96
0.33
Make
Only
7.11
1.32
Make
and Buy
2.45
0.66
Process only innovators (non-product innovators)
0.00
13.33
0.00
Organizational change innovator (non-product innovators)
8.11
13.51
2.70
Product innovators
7.41
33.33
12.96
Source: Authors’ elaboration based on Innovation Surveys
Table 23 presents the OLS regression for total employment growth for all and small
firms, and for high- and low-tech industries. The main result of these estimations is that inhouse innovation (make only) is positive and significantly associated with employment
growth. In contrast, the buy only strategy and both strategies together are not found to be
significantly associated with employment. This result holds across the different samples,
suggesting that there are no relevant differences by firm size or technological intensity of
industries
33
Table 23. Innovation Strategies
Sector
Regression
Small
Manuf
OLS
Small
Manuf
OLS
4.466***
3.122***
4.233***
3.570***
4.221***
All Manuf All Manuf
OLS
OLS
High Tech High Tech Low Tech Low Tech
OLS
OLS
OLS
OLS
Constant
3.695***
(0.757)
(1.095)
(1.101)
(1.52)
(0.904)
(1.353)
(1.374)
(1.881)
Make (dummy)
5.485***
5.923***
6.039
6.672*
4.033*
4.264*
10.225**
10.042**
(2.067)
(2.112)
(3.749)
(3.862)
(2.303)
(2.377)
(4.52)
(4.402)
1.724
1.817
4.48
4.351
-1.741
-1.796
7.597
7.327
(3.389)
(3.415)
(11.485)
(11.381)
(3.666)
(3.685)
(6.431)
(6.431)
Buy (dummy)
Make & Buy (dummy)
Foreign Ownsership (dummy)
Located in the capital region
(dummy)
3.971***
5.070***
2.287
4.804
0.433
6.771
(2.238)
(5.588)
(2.236)
(4.754)
0.596
3.972
-2.202
11.059*
(2.18)
(6.302)
(2.025)
(6.559)
-2.460*
-3.417
-1.004
-5.135*
(1.491)
(2.25)
(1.748)
(2.825)
2-digit industry dummies
Yes
yes
yes
yes
yes
yes
yes
R-squared
0.027
0.031
0.034
0.042
0.013
0.015
0.053
Number of firms
835
835
408
408
550
550
285
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent..
yes
0.083
285
The estimated impact of innovation strategies on skilled and unskilled workers is
shown in Table 24 for all and small firms and for high- and low-tech industries. For the
whole sample of firms, we find a positive and significant effect of in-house innovation over
employment growth of both types of workers. In addition, there is some evidence of
reinforcing effects of both strategies in the case of skilled workers. For the sample of small
firms, we find some differences compared to the whole sample. The make only strategy is
positive and significant for skilled workers only, and we do not find evidence of reinforcing
effects of both strategies.
In the case of high- and low-tech industries, we find that the estimated model does not
reflect the behavior of firms in high-tech industries. Most of the parameters are not
significant for that sample. However, the results for low-tech industries show that in-house
innovation positively and significantly affects employment growth for unskilled workers.
In sum, our results show that the external innovation strategy does not affect overall
employment growth either for skilled or unskilled workers. This evidence is robust across
firm size and technological intensity. In contrast, make only innovation positively affects
employment growth, but there are significant differences across types of worker, firm size,
and technological intensity.
34
Table 24. Innovation Strategies for Different Types of Labor
Sector
Regression
Constant
All
Skilled
OLS
Buy (dummy)
(1.543)
(1.072)
(1.529)
Small
Unskilled
Small
Unskilled
H-T
Skilled
H-T
Skilled
H-T
Unskilled
H-T
Unskilled
L-T
Skilled
L-T
Skilled
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
OLS
1.547
2.45
3.436**
5.162***
3.386***
3.528*
4.423***
5.667***
2.318
4.421
3.096***
4.06
(1.505)
(2.054)
(1.493)
(1.994)
(1.296)
(1.862)
(1.327)
(1.872)
(2.037)
(2.784)
(1.826)
(2.639)
14.198** 15.055***
L-T
L-T
Unskilled Unskilled
5.811*
5.669*
6.030*
1.289
2.168
2.608
3.375
3.554
3.955
(3.090)
(3.106)
(3.142)
(5.680)
(5.743)
(5.445)
(5.601)
(3.547)
(3.63)
(3.635)
(3.682)
(5.594)
(5.647)
(5.623)
(5.61)
-0.774
-0.378
2.289
2.225
-9.696
-9.595
11.64
11.427
-4.363
-4.021
-0.658
-0.829
5.357
5.91
7.344
6.379
(3.690)
(3.661)
(4.762)
(4.834)
(9.898)
(9.697)
(12.081)
(11.913)
(4.608)
(4.625)
(5.905)
(6.014)
(5.896)
(5.855)
(7.887)
(7.732)
Located in the capital
region (dummy)
Number of firms
OLS
Small
Skilled
4.823
Foreign ownership
(dummy)
R-squared
OLS
Small
Skilled
(3.042)
Make & Buy (dummy)
2-digit industry
dummies
OLS
3.005*** 3.742** 3.955*** 5.119***
(1.10)
Make (dummy)
All
All
All
Skilled Unskilled Unskilled
12.173** 12.710** 12.719**
11.875**
6.875*
1.268
10.415
3.77
5.154
0.597
10.378
4.137
(3.906)
(3.593)
(6.941)
(6.064)
(3.714)
(4.255)
(8.556)
(6.624)
2.981
0.992
5.97
7.22
1.562
-2.899
7.632
15.205*
(3.304)
(3.414)
(8.066)
(6.103)
(3.436)
(3.674)
(8.836)
(7.819)
-4.066*
-3.179
-3.529
-5.201*
-2.118
-2.232
-7.817*
-4.631
(2.267)
(2.125)
(3.148)
(2.988)
(2.688)
(2.642)
(4.209)
(3.572)
yes
yes
yes
yes
yes
yes
yes
yes
Yes
yes
Yes
yes
yes
yes
yes
yes
0.023
0.032
0.027
0.030
0.069
0.078
0.038
0.049
0.006
0.01
0.013
0.016
0.054
0.076
0.059
0.079
835
835
835
835
408
408
408
408
550
550
550
550
285
285
285
285
Notes: (1) Robust standard errors in parentheses, (2) Significance level: *** 1 percent, ** 5 percent, and * 10 percent.
7. Conclusions
In this paper we have drawn on several sources of information to shed light on the
relationship between innovation and employment growth. This quantitative approach has
been complemented by case studies and interviews of researchers and policymakers. The
overarching conclusions are that process innovation is generally found not to be a relevant
determinant of employment growth in Chilean manufacturing plants, and that product
innovation is frequently positively associated with an expansion of employment. We find
that this evidence is similar for small firms.
Our results are consistent with previous empirical evidence showing ambiguous
expected effects of process innovation and, generally, positive effects of product
innovation. This is also consistent with our discussions with researchers, policymakers, and
firm managers. Most of them agree with the expected differences in the impact of both
types of innovation. Moreover, the potential negative effect of process innovation does not
seem to be an important consideration when undertaking innovation within the firm or
implementing public innovation policies.
We also explored the impact of innovation on the composition of the workforce. As
suggested by the theoretical and empirical literature, depending on the skill bias of
innovation, its impact may be different for skilled and unskilled workers. We find that
process innovation, in general, has a negative and significant effect on unskilled
employment growth, with the exception of low-tech industries. By contrast, our findings
show that product innovation increases both types of employment growth. The
decomposition of employment growth in its main components reveals some interesting
results. For skilled workers, the net contribution of product innovation is positive and large
for all firms and those in high-tech industries. In the case of unskilled workers, we find that
the net contribution of product innovation is generally negative. Thus, these results suggest
some evidence consistent with the idea that product innovation would be skill biased in the
Chilean case. However, these results, based on OLS regressions, should be viewed with
caution, given that they are not confirmed by estimations with instrumental variables.
Finally, using information for one innovation survey only, we analyzed the effect of
two different innovation strategies on employment growth and skill composition. Our
results show that in-house innovation (make only) is positive and significantly associated
with employment growth. In contrast, buy only and make-or-buy strategies are not found to
be significantly associated with employment growth. These findings hold across all
samples, suggesting that there are no relevant differences with respect to firm size or
technological intensity.
The effect of innovation strategies on skill composition is more mixed. In general,
external innovation does not affect employment growth of skilled and unskilled workers.
This evidence is robust across firm size and technological intensity of industries. In
contrast, the effect of “in house” innovation is different across types of workers, firm size,
and technological intensity. We found that R&D within firms increases employment growth
of skilled workers in small firms, but it has no impact on unskilled workers in these firms.
In addition, any of these strategies affects employment growth of skilled workers in hightech industries, but “in house” R&D increases employment growth for both types of
workers in low-tech industries. These mixed results may be attributable to the broad
categories that are used in the estimations. There is a lot of heterogeneity within both
categories affecting the chances of finding reasonable results. It reveals that more
disaggregation is potentially required to investigate the negative effects of technological
change on unskilled workers.
These findings have some relevant policy implications. First, given that product
innovation and process innovations have similar effects on large and small firms, public
policy would not need to be concerned with the effects of innovation on employment for
small firms. In fact, a first order issue is how to make small firms innovate more and take
advantage of the positive effects of innovation on employment growth.
Second, our finding of a negative effect of innovation on unskilled workers calls for
further analysis. Given that innovation can displace less skilled workers, a closer interaction
between innovation and training policies in Chile might be required. As far as we know as a
result of our interviews with managers and policymakers, there are no specific public
programs dealing with this issue. Although we do not address labor marker regulations
specifically, this evidence may be consistent with the idea that raising the cost of hiring
unskilled workers, as a result of minimum wages increases, may bias innovation against
less qualified workers. This is an issue that need be addressed empirically, but
policymakers need to be aware of this potential collateral issue related to labor regulations.
37
Finally, the differences in the effect of in-house R&D on low- and high-tech industries
can have some implication for the recent reformulation of the R&D tax credit. Whenever
this subsidy incentivizes internal R&D, there may be differential effects on employment
growth for high- and low-tech industries. According to our results, in-house R&D tends to
increase employment for both types of workers in low-tech manufacturing industries. The
effect on high-tech industries is an issue deserving more attention by policymakers in Chile.
38
References
Acemoglu, D. 1998. “Why Do New Technologies Complement Skills? Directed
Technical Change and Wage Inequality.” Quarterly Journal of Economics
113: 1055-1089.
Alvarez, R. 2001. “External Sources of Technological Innovation in Chilean
Manufacturing Industry.” Estudios de Economia 28(1): 53–68.
Alvarez, R., Bravo-Ortega, C. and L. Navarro (2010): “Innovation, R&D
Investment and Productivity in Chile.” IDB Working Paper Series No. IDBWP-190. Washington, DC, United States: Inter-American Development
Bank.
Alvarez, R. and R. Robertson. 2004. “Exposure to Foreign Markets and FirmLevel Innovation: Evidence from Chile and Mexico.” Journal of
International Trade and Economic Development 13 (1): 57–87.
Benavente, J.M. 2006 “The Role of Research and Innovation in Promoting
Productivity in Chile.” Economics of Innovation and New Technology 15
(4-5): 301–315.
Benavente, J.M. and R. Lauterbach. 2008. “Technological Innovation and
Employment: Complements or Substitutes?” The European Journal of
Development Research 20(2): 318–329.
Berman. E., J. Bound, and Z. Griliches. 1994. "Changes in the Demand for
Skilled Labor within U.S. Manufacturing Industries: Evidence from the
Annual Survey of Manufacturing." NBER Working Papers 4255.
Cambridge, MA, United States: National Bureau of Economic Research.
39
Bergoeing, R., A. Hernando, and A. Repetto. 2005. “Deflators and Capital
Series for the ENIA: Technical Appendix.” Centro de Economía Aplicada,
Universidad de Chile. Unpublished.
Brandt, N. 2010. “Chile: Climbing on Giants' Shoulders: Better Schools for all
Chilean Children.” OECD Economics Department Working Papers No. 784.
Card, D. and J. E. DiNardo. 2002. “Skill Biased Technical Change and Rising
Wage Inequality: Some Problems and Puzzles.” Journal of Labor
Economics 20(4): 733–783.
Crespi, G. 1999. “Determinantes de la Innovación Tecnológica en la Industria
Manufacturera Chilena.” Estadistica y Economía 17: 113–34.
Crespi, G. and J. Katz. 1999. “R&D Expenditure, Market Structure and
Technological Regimes in Chilean Manufacturing Industry.” Estudios de
Economia 26(2): 163–86.
Crespi, G. and P. Zuñiga. 2010. “Innovation and Productivity: Evidence from
Six Latin American Countries.” IDB Working Paper No. 218. Washington,
DC, United States: Inter-American Development Bank.
Dachs, B. and B. Peters 2011. “Innovation, Employment Growth and Foreign
Ownership of Firms: A European Perspective.” Austrian Institute of
Technology and Center for European Economic Research. Unpublished.
Fuentes, J. R., M. Larraín, and K. Schmidt-Hebbel. 2006. “Measuring and
Explaining Total Factor Productivity in Chile.” Cuadernos de Economía,
43(May): 113–42.
Gallego, F. 2011. “Skill Premium in Chile: Studying Skill Upgrading in the
South.” Documento de Trabajo IE-PUC, N° 396.
40
Galia, F. and D. Legros. 2004. “Complementarities between Obstacles to
Innovation: Evidence from France.” Research Policy 33: 1185–99.
Hall, B. H., F. Lotti, and J. Mairesse. 2008. “Employment, Innovation, and
Productivity: Evidence from Italian Microdata.” Industrial and Corporate
Change 17(4): 813–39.
Harrison, R., J Jaumandreu, J. Mairesse, and B. Peters. 2008. “Does Innovation
Stimulate Employment? A Firm Level Analysis using Comparable MicroData from Four European Countries, NBER Working Paper 14216.
Cambridge, MA: National Bureau of Economic Research.
Iammarino, S., F. Sanna-Randaccio, and M. Savona. 2007. "The Perception of
Obstacles to Innovation. Multinational and Domestic Firms in Italy."
Working Papers BETA 2007-12. Strasbourg, France: Bureau d'Economie
Théorique et Appliquée, UDS.
Kharas, H., D. Leipziger, W. Maloney, et al. 2008. “Chilean Growth through
East Asian Eyes.” Working Paper No. 31, Economic Growth Commission.
Maloney, W. and A. Rodríguez-Clare. 2007. “Innovation Shortfalls.” Review of
Development Economics 11(4): 665–84.
Mohnen, P., F. Palm, S. S. van der Loeff, and A. Tiwari. 2008). “Financial
Constraints and Other Obstacles: Are they a Threat to Innovation Activity?”
The Economist 156(2): 201–14.
OECD. 1997. “Oslo Manual: Proposed Guidelines for Collecting and
Interpreting Technological Innovation Data.” Paris, France: OECD.
41
Peters, B. 2005. "Employment Effects of Different Innovation Activities:
Microeconometric Evidence." ZEW Discussion Papers 04-73. Manheim,
Germany: Center for European Economic Research.
Pianta. M. 2005. "Innovation and Employment." In J. Fagerberg, D. Mowery
and R. Nelson, editors. The Oxford Handbook of Innovation. Oxford,
England: Oxford University Press.
Savignac, F. 2008. "Impact of Financial Constraints on Innovation: What Can
Be Learned From A Direct Measure?" Economics of Innovation and New
Technology 17(6): 553–69.
42
Appendix 1
Case Studies of Chilean Firms
Case 1: Cuerotexsa
Cuerotexsa is a small business devoted to manufacturing and marketing synthetic leather
products. It was founded in 1973 to meet the needs of new markets, especially in LatinAmerica. Since its founding, the company has been dedicated to developing a service
orientation, which has enabled it to survive in a highly competitive market, especially in the
face of competition from China.
The firm’s main export market is the Argentinean women’s shoe market, and it has
been forced to innovate constantly in order to compete. According to the general manager,
Cuerotexsa has had to develop several innovations in its productive process in order to
increase the efficiency of production. The implementation of process innovation has been a
response to requirements to reduce pollutants emission and adjust to environmental
regulations, increase production speed, and reduce the number of days that the factory has
to close.
The main source of innovation in Cuerotexsa has been external, the result of
acquisition of new machinery for improving its productive process. There is little “inhouse” innovation. However, the company has developed its own innovation in the
production of synthetic leather aimed at the Argentinean women’s shoe market, because of
the need to adjust to new trends in the fashion industry.
Carlos Jofré, manager of Cuerotexsa, pointed out that the company has undergone a
cultural shift through significant changes in the workforce. During the last three years,
employment has maintained relatively stable, increasing slightly from 41 workers in 2007
to 45 in 2010. He underscored the close relationship between the type of innovation and
employment generation, not with innovation itself. Jofré believes that product innovation is
more closely associated with increases in employment. In contrast, process innovation,
whose aim is to improve efficiency, can have a negative effect on employment.
Cuerotexsa considers that obstacles to innovation are more related to a lack of
resources than to internal characteristics of the business. Both the workers and management
are aware that in order to survive and keep growing, innovation efforts must continue.
43
An analysis of the government’s innovation policies points shows that they have been
implemented with the correct and most relevant objectives, but that there is some positive
bias toward firms that have experience in the application of public instruments. The
absence of clear and accessible information is one of the biggest obstacles that the small
and medium sized enterprises must overcome in order to access the government’s proinnovation policies. Finally, this firm highlights the existence of high transaction costs
compared with the benefits that public programs could bring to firms.
Case 2: SlipNaxos
SlipNaxos Chile S.A. was founded in 1997 from the cooperation between Chilean
entrepreneurs and the Swedish firm NaxoFlex AB. It focuses on providing solutions in the
abrasive and adhesive area to the timber and the metalworking industries. SlipNaxos is a
small enterprise employing 18 workers. It specializes in the commercialization of abrasive
and adhesive products. In the last three years, employment has grown by about 30 percent,
explained by the import of new products and the increase in domestic production due to
improvements in the manufacturing process of abrasive products.
The company considers itself open to new possibilities and innovation. Although it
believes that the lack of resources for small companies decreases or delays innovation
SlipNaxos has been able to develop in-house innovation for marketing its products. In
contrast, most of the product innovation comes from its international partners.
Christian Enriquez, SlipNaxos’ manager, points out that public funds aimed at
encouraging innovation are more targeted to the creation of new technologies than to their
adoption. This is only one of the many obstacles that small and medium sized companies
have to overcome in order to access these funds. In addition, the lack of information and the
scarce correlation between expected gains and the cost are issues that Enriquez sees as
fundamental problems to be resolved.
Enriquez also points out that there is a relationship between the type of innovation and
employment growth. He believes that a product innovation should bring about an increase
in employment, as occurred in this firm with the import of new products of the American
brand “Franklin Titebon”. In the case of process innovation, he considers that it should
generate a decrease in employment. In his own experience, this has not been the case.
44
SlipNaxos has been able to increase workers’ training and maintain the same number of
employees.
Case 3: CMPC Tissue
With factories in Chile, Argentina, Peru, and Uruguay, CMPC Tissue manufactures and
markets tissue products such as toilet paper, kitchen paper towels, and tissue paper based on
the use of cellulose and recycled paper. CMPC Tissue is considered the leader in the
Chilean, Argentinean, and Uruguayan tissue market, and in Peru its market share is
growing constantly and currently is in second place. In these countries, CMPC also
manufactures specialized products for enterprises, institutions, and external clients with a
marketing structure especially designed for these costumers.
Eduardo Lucachewski, manager of the plant located in Santiago, said that the company
is in a constant process of innovation in both products and processes. He asserts that most
innovations in this firm have been the result of strategic alliances with various machinery
producers for the paper industry. These partners conduct various kinds of research and are
financed by CMPC. In this sense, this company produces external innovation, but with a
high degree of involvement.
The manager remarked that innovation in this company has not greatly affected the
level of employment. However, he thinks that innovation generates two types of effects: (i)
niche product innovation generates an increase in employment because it increases the
demand for goods produced by the firm, and (ii) process innovation, by contrast, results in
substitution between the use of machinery and labor, decreasing the demand for workers.
However, he noted that in the last three years, the product innovation effect has
predominated, and CMPC Tissue has experienced a slight increase in the number of
employees.
Lucachewski also pointed out the differences between in-house and external
innovation. He believes that in-house innovation would probably generate a substitution
effect between unskilled and skilled labor (researchers) and a productivity effect that will
slow down employment, versus external innovation, which produces only a decrease in
unskilled labor.
45
Finally, the manager evaluated the government’s innovation policies, concluding that
the biggest weakness in Chilean innovation policy is the lack of incentives and specific
measures for facilitating worker training. He proposed the creation of training programs
tailored to the needs of each company, according to the number of workers and the time
availability of the company.
Case 4: CALPER
CALPER was founded in 1990 to meet a latent demand in the work shoes market. This firm
is specialized in the development, manufacturing, and commercialization of occupational
shoes, using Chilean leather and high-quality inputs imported from Brazil and Mexico. This
is a dynamic firm that has been able to compete successfully with imports from low-wage
countries such as China. Since 2007, it has experienced an employment growth of about 90
percent, explained, according to the marketing manager, by an increase in sales and product
innovation.
Gonzalo Perez Urquiaga, marketing manager of the firm, pointed out that Calper is a
firm that constantly innovates in the areas of marketing, product, management, and process.
The main reason behind these broad innovation efforts is increasing competition from
China’s low-cost products. In this case, innovation is considered a condition for the firm’s
survival.
This company is an interesting example of in-house innovation. It had to develop and
manufacture its own machinery for producing shoes, given that the market for this type of
machinery lacks suitable products for the firm’s needs. Then, to meet the demands of the
market and to provide better products, Calper was forced to develop and innovate in-house.
Regarding the employment effects of innovation, Perez considers them to be broadly
positive. However, he thinks that successful experiences need the joint collaboration of
workers and managers; otherwise, innovation efforts are destined to failure.
The expressed a high degree of satisfaction with the technological platform for public
purchases (“ChileCompra”), which allows this firm to sell its products to government
agencies. However, he is disappointed with the absence of protection of domestic shoes
producers and the decreasing incentives to improve workers’ capacity through educational
programs.
46